| The analysis of vibration signals has proven to be a powerful and effective tool for the early detection of developing failure in rotating machines. A variety of signal processing techniques has been used for this purpose but a complete consensus on their effectiveness does not yet exist. Although the traditional techniques are useful in simple cases and may provide a prediagnostic in the complex machine but they have several limitations which reduce their performance.;The analysis of non-stationary, cylco-stationary or time-varying signals needs advance methods capable to represent signals in time-frequency-amplitude domain. The objective of the present work is to study, compare, modify and adapt the time-frequency representation techniques which can be used to analyses non-stationary phenomena.;In the first part of this thesis the classical vibration techniques in time domain and in frequency domain are presented and the advantages and disadvantages of each technique are described. In fact, in certain cases such as transient events in a machine or varying speed rotating machinery, traditional vibration analysis methods in time or in frequency are incapable of reflecting changes in the operating conditions of machines. Time-frequency methods are introduced to solve some of these problems. Among these methods, Short-Time Fourier Transform (STFT) is considered to be the simplest technique of analysis.;In the second part of this thesis, some of the time-frequency methods such as Wigner distribution, Choi-Williams distribution, and RID distribution are briefly reviewed and advantages and disadvantages of different methods are examined. Then the capacity of each method is examined in a practical application by an in-house program developed for all of time-frequency methods.;The third part of this research deals with the development of new techniques such as wavelet analysis, by which it is possible to compensate for weaknesses in other time-frequency methods. Wavelet analysis has a special characteristic of time-frequency localization, which is very effective in the analysis of transient or time-varying signals.;The fourth part of this thesis presents a user-friendly software designed to be a bridge between theoretical research in time-frequency/time-scale methods and the practical applications of these methods in different domains. (Abstract shortened by UMI.)... |
